Jazzfan1971 and FloppyMoose's Season Preview - The Warriors

[jazzfan1971]

Golden State Warriors

Projected Lineup
-------------------------------

PG - Stephen Curry, Shawn Livingston, Leandro Barbosa, Nemanja Nedovic
SG - Klay Thompson, Brandon Rush
SF - Andre Iguodala, Harrison Barnes
PF - David Lee, Draymond Green, Marreese Speights
C - Andrew Bogut, Festus Ezeli, Ognjen Kuzmic

Trading Block: I'm not sure I see a whole lot of action on the trade front here. We are getting into teams that are looking to compete for a title now, and those teams generally don't like to shake things up a whole lot unless absolutely necessary. If they can find a way to cheaply add another depth piece or two I'm sure that they'll jump on it though. Oh, and if some team wants to make a reasonable offer for Lee I'd imagine that they'd be happy to sit and talk.

Position Battle: Could Draymond Green challenge David Lee for a starting spot? I think he could. The Ws could use a little more offense off the bench and Lee could provide that. I would still expect Lee to get nearly the same amount of minutes even if moved to the bench, so not a huge change.

Mystery Man: Gotta be Steve Kerr. Personally I never understood the hype for Kerr as a coach. It all seemed like it was served up on a silver platter to me. Maybe he'll live up to the hype, maybe he'll live up to his payday, maybe he'll have a career path much like Mark Jackson. Time will tell. On the plus side, the Warriors added a couple excellent assistants in Gentry and Adams, that alone should help ensure that Kerr looks good in his first year.

Floppymusings: First Jazzfan invites me to do the record projections for this series of posts, and then he talks down my approach (see below). I'm having an Admiral Akbar moment here... "It's a trap!!!" I'll comment on that more later in the thread. For now let's discuss GSW.

So... Is GSW a great team? Or is it the greatest team ever? What's that you say? Rookie coach? No D PF? Shallow bench? A bust on their last draft pick? A center collection that combined for well under 2000 minutes total last year? Fine, I'll put you down for "great".

Jazzfanramblings: Our projections this year are based largely on advanced stats. Namely, RAPM or Real Adjusted Plus Minus. I'm not a math major and when I took stats in college I took the lowest level they offered and just payed enough attention to pass the stupid class. Now I kinda wish I had payed more attention to the lecture and less to the brunette a couple rows down because I'd have a shot at explaining what a Bayesian technique is or how it works. But, I honestly can't. It's beyond me.

There was a time in the NBA when General Managers would scoff at stats. You'd hear them say things like, "That's all nice, but, I believe what my eyes tell me." Those days are over. Today every team is either invested in stats nerds or scrambling to become invested in stat nerds. I have trouble imagining a GM come out today and say he prefers to believe what his eyes tell him over what the nerds with the sharp pencils are telling him from their cubicle in the basement.

But, I'm old school. I always prefer to use my eyes over a bunch of stats. Stats are nice and all, but, I often feel like when folks use them they present them as indicating something slightly different than what the stat actually indicates. I mean, you have to really pay attention when discussing stats otherwise you can be easily misled. Heck, even when paying attention you can be easily misled.

So, what do my eyes tell me? Well, my eyes see Larry Sanders as the best defender in the league. My eyes see Ricky Rubio as a great PG. My eyes see Kevin Love as a middle of the road PF. In short, my eyes view the league a LOT differently than just about everyone else. But, I still trust them. They are MY eyes after all.

What I have trouble trusting is advanced stats like RAPM. Stats that tell me Andre Iguodala is the 3rd best difference maker in the league. Amir Johnson is the 6th. Nick Collison is the 12th. These things just don't pass the sniff test for me. Maybe they do for you. Maybe that is much less outlandish than my opinions on Larry Sanders or Ricky Rubio. That's fine. Reasonable people can disagree.

If I was the owner of a team I think I'd be looking for the GM with the best pair of eyes though. A guy who consistently has a track record of finding good players based solely on his observations and not at all on spreadsheets. You can have a spreadsheet guy as well, but, as support for your GM not to replace him. I still believe in the power of the human brain to be able to spot talent and weed out all of the other factors better than ANY advanced stat.

Your mileage may vary.

On the Warriors, I like their team if healthy. I could even see them coming out of the West if everything breaks their way. The West isn't wide open, but, I think 6 or even 7 teams have a shot of coming out of the West this year, and the Warriors are in that class.

Projected Record - 52/30

[Rosque]

So...they'll barely make the playoffs?

[floppymoose]

So first off, this is a big experiment. I'm trying out an approach to predict team quality that is new to me. I've certainly been influenced by rapm values for player before in my own estimates of team records, and even by other, earlier and less refined plus-minus values. In general I think exposure to that data has been a huge plus.

Adjusted plus-minus, when it first appeared publicly for nba players, was the first stat to agree with my eyes on a lot of players who put up big box score numbers but actually sucked. Like Troy Murphy or Michael Beasley.

Jazzfan told me that he chose me to do the projections because I had identified problems with his projections in last year's thread. Those included his predictions that:
the Raptors would miss the playoffs, and be worse than both the Pistons and the Bucks
the Nets would be the best team in the league
the Pistons would win 41 games
the Bucks would win 33 games and be better than 8 other teams
that other players in Denver would step up and replace a lot of Iggy's contribution
that Byron Mullens was going to get minutes with the Clippers and was a fantastic addition

I actually don't expect my methods to fare very well in this first attempt. I spent a *lot* of time trying to get the calculations set up this year, and of dealing with imperfect data sources. That has robbed me of some time I would have liked to have to tune the expected minutes played values for all the players. I also expect to make progress in future years in doming some prediction of player improvement or decline, and of smoothing out outliers in player values if we have data from other recent years for that player. Basically there is a long way to go here.

But jazzfan's eyes are far from magical and are not setting a high bar. I don't think it will take many seasons for the steam drill to catch John Henry.

[bondom34]

So first off, this is a big experiment. I'm trying out an approach to predict team quality that is new to me. I've certainly been influenced by rapm values for player before in my own estimates of team records, and even by other, earlier and less refined plus-minus values. In general I think exposure to that data has been a huge plus.

Adjusted plus-minus, when it first appeared publicly for nba players, was the first stat to agree with my eyes on a lot of players who put up big box score numbers but actually sucked. Like Troy Murphy or Michael Beasley.

Jazzfan told me that he chose me to do the projections because I had identified problems with his projections in last year's thread. Those included his predictions that:
the Raptors would miss the playoffs, and be worse than both the Pistons and the Bucks
the Nets would be the best team in the league
the Pistons would win 41 games
the Bucks would win 33 games and be better than 8 other teams
that other players in Denver would step up and replace a lot of Iggy's contribution
that Byron Mullens was going to get minutes with the Clippers and was a fantastic addition

Wait.....to clarify these points further:


the Raptors would miss the playoffs, and be worse than both the Pistons and the Bucks - They traded the highest usage players on their roster and essentially completely changed the team's offensive and defensive dynamics. The team post Gay and Bargs trades played much better than pre-trade. So, really projections couldn't account for the trade.


the Nets would be the best team in the league - Many people said this, but didn't account for age, injury, and a new head coach. Again, you're not projecting injuries and coaching.

the Pistons would win 41 games - Josh Smith is still not a SF, but that's what 2 coaches were fired for.

the Bucks would win 33 games and be better than 8 other teams - The Bucks lost I don't even know how many games to injury. They lost Sanders, Zaza, Delfino, and Ersan I believe as well as some others. Again, projections wouldn't account for this.

that other players in Denver would step up and replace a lot of Iggy's contribution - And again, Gallo, McGee, and others were injured.


that Byron Mullens was going to get minutes with the Clippers and was a fantastic addition - The hell?


So looking at these, RAPM won't help at all. I'm a proponent of its usefulness, but for an exercise like this you need minutes projections, and I'd guarantee the projections for these teams were way off due to injury and trades. The fact that Detroit played a talented player at the wrong position wasn't going to be accounted for either. I have no idea about the Mullens thing.

[jazzfan1971]

No, my eyes were pretty bad with my predictions. There is a reason I brought in Floppy. I have faith him him.

I think RAPM is actually going to do a better job at estimating team success than it does player ranking. There is just a lot more data to work with when you are dealing with 8 or 9 guys vs. 1.

Injuries can explain a lot of my errors away, true. But, I assume that this time next year we'll be using injuries to explain away a lot of Floppy's errors. You can't really use any system to predict injuries.

What is going to be most impressive in my opinion is how close his overall ranking of teams from worst to first lines up. Not so much the win totals. I'll also be impressed if a few of his more controversial predictions ring true. That IMHO will really validate his method.

And Mullins, I am sure I was talking about Chris Mullin, wasn't I?

[floppymoose]

So looking at these, RAPM won't help at all.

That's not really true. I absolutely would have predicted less minutes for the Net's ancient cast than they played the year before. Their age, injury history, and minutes trend should all affect their projections. RAPM values were also why I thought Iggy would be hard to replace, and why I knew Mullens wasn't going to play on a very good Clippers team that had much better options. In Detroit's case they weren't just gaining Smith. They were gaining Jennings and thrusting him into a big minutes role. Jennings was a big rapm hole. I will admit that the Bucks bit wasn't based on anything but my own instincts. I just felt they didn't have a real PG anywhere on the team and that would spell disaster. In the end Sanders missing all that time and Mayo turning into a pumpkin were more important.

It's nice to hear jazzfan has faith in me but I don't really have faith in either of us. I guess I'm a glass half empty type. But I do think it will be both fun and a good learning exercise to see how good a job can be done by a quantitative approach to predicting team performance.

[bondom34]

So looking at these, RAPM won't help at all.

That's not really true. I absolutely would have predicted less minutes for the Net's ancient cast than they played the year before. Their age, injury history, and minutes trend should all affect their projections. RAPM values were also why I thought Iggy would be hard to replace, and why I knew Mullens wasn't going to play on a very good Clippers team that had much better options. In Detroit's case they weren't just gaining Smith. They were gaining Jennings and thrusting him into a big minutes role. Jennings was a big rapm hole. I will admit that the Bucks bit wasn't based on anything but my own instincts. I just felt they didn't have a real PG anywhere on the team and that would spell disaster. In the end Sanders missing all that time and Mayo turning into a pumpkin were more important.

MKE and the Raps would have been impossible. Denver would have been at least helped by Gallo, and I think some of Jennings would be offset by Smith. As for the Nets, who else was going to get minutes, the entire team was old? RAPM would have predicted a great season. I'm pretty sure nbacouchside made posts here last year that projected them very well using xRAPM, which has box score added to it but is still similar enough.

[floppymoose]

Jennings would absolutely be offset by Smith. And that's pretty much what I predicted: that the Pistons would be about as bad as they were the year before. I don't really understand your Nets comment. Obviously you predict backups and replacement players would get those minutes, and it should come as no surprise that backups and replacement players have worse rapm values than Deron, KG, Pierce, Lopez, and AK47.

edit: oh, and the Raps were replacing Bargnani with *anyone else*. That was why I thought they'd improve. I actually only thought they'd sneak in as an 8th seed. They surprised me. But losing the Bargs rapm destroyer was absolutely why I thought they'd be better.

[bondom34]

Jennings would absolutely be offset by Smith. And that's pretty much what I predicted: that the Pistons would be about as bad as they were the year before. I don't really understand your Nets comment. Obviously you predict backups and replacement players would get those minutes, and it should come as no surprise that backups and replacement players have worse rapm values than Deron, KG, Pierce, Lopez, and AK47.

edit: oh, and the Raps were replacing Bargnani with *anyone else*. That was why I thought they'd improve. I actually only thought they'd sneak in as an 8th seed. They surprised me. But losing the Bargs rapm destroyer was absolutely why I thought they'd be better.

Here is what I was referring to w/ xRAPM, it was blended plus/minus:
http://nbacouchside.com/2013/10/29/one- ... nt-method/

Nets with 53 wins, Detroit with 36. The Nets issue to me is that it wasn't necessarily the minutes distribution, a ot was age and coaching. JJ, Deron, and KG played a ton ,but 2 of the 3 were not very good

[floppymoose]

Not a believer in xRAPM, for what it's worth. It came in last of the three rapm flavors I tried when simulating last season's records with last seasons actual player minutes and actual rapm values.

The post you link to does not show the player minutes he estimated. Or what he did for very low minute players, and what kind of replacement player values he used. That is likely the source to the difference.

Although I should also say that I thought the the Nets would be better then they were. I just didn't think they would be 1st, or close to it.

[bondom34]

Fair enough, and I'm not saying they won't be an improvement, just saying there will still be unpredicable results in the end. There always are, that's why they play the games.

[floppymoose]

Absolutely. That's why none of ESPN's 18 experts got more than 6 of the 8 first round playoff series correct two years ago.

I don't expect my way to be super accurate either. The real goal this first year is to do better than average compared to human competition.

[timlin500]

I think Warriors can win about 55 games this year. Liking the new offense that's been implemented so far, and the bench depth is so much better than last year's.

[Hoopstar23]

I think Warriors can win about 55 games this year. Liking the new offense that's been implemented so far, and the bench depth is so much better than last year's.

That offense looks really scary, the health of the warriors and the improvement of Klay Thompson should have them as a WCF birth or even them coming out of the West with a shot to win it all

[JimmyTD3]

Jazzfan: I agree that the Warriors will win about 52-55 games.

But, about 65% of your post was you stroking yourself, and had nothing to do with the Warriors.

Since they are my team, I would've liked a little more analysis on...well...the actual team....and not about your eyes, how you're old school, and your thoughts on advanced stats.

[timlin500]

Jazzfan: I agree that the Warriors will win about 52-55 games.

But, about 65% of your post was you stroking yourself, and had nothing to do with the Warriors.

Since they are my team, I would've liked a little more analysis on...well...the actual team....and not about your eyes, how you're old school, and your thoughts on advanced stats.

LOL. I concur

[jazzfan1971]

Eh, that's how all my posts have been.

The thing of it is, most fans of GS know GS better than I do. And that stands for every team. So, it really wouldn't be that helpful for me to try and educate fans on their own teams. So, I don't really try to go there. I leave that for fans of the team to fill in here in the thread as feedback. And sometimes I get great feedback on teams and sometimes I don't.

What I try to do is spark debate. My thoughts here on advanced stats vs. eyes was intended to launch a debate about well... advanced stats vs. eyes. Obviously it hasn't done much on that front. But, the intention was there. You can usually find an intention like that with all my Jazzfanramblings. I try to put forward a thought of some sort that is at least a little bit controversial and hope that it leads to a discussion. If I come across some bit of information that I feel would be interesting to general NBA fans I also try to include that. (for instance I don't know that a general NBA fan would know that the Wiz are targeting Durant in a couple years, so I include that, not for Wizard fans who probably are well aware, but, for the fans of other teams who may or may not have that information)

If you want more analysis on your team, go ahead and provide it. I'm quite sure you could do a better job on the Warriors than I could and there would no doubt be folks that would appreciate your contribution to the thread.

[Onus]

I think the Warriors are going to be pushing 55+ wins this year. This new offense should push us to be a top 5 offense and if we can maintain our top 5 defense, darkhorse bet.

[floppymoose]

Part of the reason that I don't have the Warriors (and some other team coming up in the previews) with more wins is that the league is getting top heavy this year. I only have 10 teams of the 30 projected to be under .500. If that ends up be correct, or even close to correct, it's going to be hard to get many teams up to 55 wins.

I do have IND with 44 wins, and MIN with 38 wins, and both of those might be pretty controversial. If those teams end up a lot worse then there will be more wins to go around to the top teams.

[snomeister]

What I have trouble trusting is advanced stats like RAPM. Stats that tell me Andre Iguodala is the 3rd best difference maker in the league. Amir Johnson is the 6th. Nick Collison is the 12th. These things just don't pass the sniff test for me. Maybe they do for you. Maybe that is much less outlandish than my opinions on Larry Sanders or Ricky Rubio. That's fine. Reasonable people can disagree.

Or maybe, these players just do the little things, that while we fans don't traditionally think of as important, actually affect the outcome of basketball games much more than we believe.

I'm a huge fan of Amir and I've long told Raptor fans of his importance to this team. He sets good screens to initiate the offense, is a much better rebounder than his raw stats suggest as he is among league leaders in contested rebounds, boxes out very well (team rebounding is considerably better when he's on the court), plays great defense in every aspect including stopping the pick and roll and challenging shots, hustles on EVERY play (even when injured), and is a consummate professional as he always looks to help his teammates even after the play and only takes good shots (among league leaders in FG% despite taking an unusual amount of 3s). And I can go on. The man does everything on the floor outside of putting up a bunch of raw stats like points. Do I think he is the 6th best player in the league? No, but he is crazy underrated (I don't really care about ESPN's player countdown but I did think it was pretty insulting to have him around 106 or something, but maybe they would have him higher using a different method of ranking), and imo he is even more important to our team than Demar Derozan.

Amir doesnt do a lot to wow you, but it's a lot harder to find an elite glue guy like Amir than find a scorer. He's probably a player that the Spurs would love to have. The eye test works but you also need to know what to look for, and Amir definitely passes it. And the same thing can be said about Iggy and Collison. I agree that stats need to be taken in context but in my experience, RAPM is fairly accurate in measuring a player's impact.